Using the K Nearest Neighbor Method and BM25 in the Patent Document Categorization Subtask at NTCIR-5
Masaki Murata, Toshiyuki Kanamaru, Tamotsu Shirado, Hitoshi Isahara · NTCIR · 2005
Patent processing is extremely important for industry, business, and law. We participated in the F-term categorization subtask at NTCIR-5, in which, we classified patent documents into their F-terms using the k-nearest neighbor method. For document classification, F-term categories are both very precise and useful. We entered five systems in the F-term categorization subtask. They obtained the best f-measures of all 18 participating systems. This confirmed the effectiveness of our method. After the contest, we performed the experiments again during the theme categorization subtask, even though we did not officially enter. The results showed that our system obtained higher f-measures than the highest obtained by the other systems performing the subtask. This also confirmed the effectiveness of our method.